U.S. patent application number 15/214561 was filed with the patent office on 2018-01-25 for evaluating temporal relevance in cognitive operations.
The applicant listed for this patent is International Business Machines Corporation. Invention is credited to Charles E. Beller, Paul J. Chase, JR., Richard L. Darden, Edward G. Katz.
Application Number | 20180025280 15/214561 |
Document ID | / |
Family ID | 60988684 |
Filed Date | 2018-01-25 |
United States Patent
Application |
20180025280 |
Kind Code |
A1 |
Beller; Charles E. ; et
al. |
January 25, 2018 |
Evaluating Temporal Relevance in Cognitive Operations
Abstract
Mechanisms are provided for evaluating a temporal relevance of a
portion of content to a cognitive operation request. A cognitive
operation request is received that comprises a portion of input
text and the input text is analyzed, by a temporal relevance
evaluation engine, to identify a temporal focus of the input text.
A corpus of content is processed based on the input portion of text
to generate candidate results each of which are processed to
identify at least one contextual temporal focus associated with the
candidate result. The at least one contextual temporal focus is
compared with the temporal focus of the input text and a measure of
temporal relevance of the candidate result is generated based on
results of the comparison. The cognitive operation is performed
based on the measure of temporal relevance.
Inventors: |
Beller; Charles E.;
(Baltimore, MD) ; Chase, JR.; Paul J.; (Fairfax,
VA) ; Darden; Richard L.; (Leesburg, VA) ;
Katz; Edward G.; (Washington, DC) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
International Business Machines Corporation |
Armonk |
NY |
US |
|
|
Family ID: |
60988684 |
Appl. No.: |
15/214561 |
Filed: |
July 20, 2016 |
Current U.S.
Class: |
706/58 |
Current CPC
Class: |
G06F 16/24522 20190101;
G06N 5/04 20130101 |
International
Class: |
G06N 5/04 20060101
G06N005/04; G06F 17/30 20060101 G06F017/30 |
Goverment Interests
[0001] This invention was made with United States Government
support under Agreement No. 2013-12101100008. THE GOVERNMENT HAS
CERTAIN RIGHTS IN THE INVENTION.
Claims
1. A method, in a data processing system comprising at least one
processor and a memory, the memory comprising instructions which
are executed by the at least one processor to specifically
configure the at least one processor to implement a cognitive
system comprising a temporal relevance evaluation engine for
evaluating a temporal relevance of a portion of content to a
cognitive operation request, the method comprising: receiving, by
the cognitive system, a cognitive operation request comprising a
portion of input text; analyzing, by the temporal relevance
evaluation engine, the input text to identify a temporal focus of
the input text; processing, by the temporal relevance evaluation
engine, a corpus of content based on the input portion of text to
generate candidate results; processing, by the temporal relevance
evaluation engine, each candidate result, to identify at least one
contextual temporal focus associated with the candidate result;
comparing, by the temporal relevance evaluation engine, the at
least one contextual temporal focus with the temporal focus of the
input text; generating, by the temporal relevance evaluation
engine, a measure of temporal relevance of the candidate result
based on results of the comparison; and performing, by the
cognitive system, the cognitive operation based on the measure of
temporal relevance.
2. The method of claim 1, wherein the input text is a natural
language question that has at least one term corresponding to a
temporal focus of the natural language question, and wherein the
candidate results are candidate answers to the natural language
question generated from documents of the corpus.
3. The method of claim 1, wherein the input text comprises at least
one search term of a search query, and wherein the at least one
search term comprises a term corresponding to a temporal focus of
the search query, and wherein the candidate results are candidate
search results of the search query generated from documents of the
corpus.
4. The method of claim 1, further comprising analyzing the
documents of the corpus, wherein analyzing the documents of the
corpus comprises, for each document: determining a document
relevance datetime of the document based on at least one of
metadata or content of the document, wherein the document relevance
datetime applies to all content of the document; and associating,
with each definite temporal expression in the content of the
document, a datetime with the definite temporal expression, and
wherein the at least one contextual temporal focus of a candidate
result generated based on a definite temporal expression in the
content of the document is identified based on at least one of the
document relevance datetime or the datetime of the definite
temporal expression.
5. The method of claim 4, wherein analyzing the documents of the
corpus further comprises, for each document: associating with each
token in the content of the document, a datetime based on a
document relevance datetime associated with the document and a
datetime associated with a definite temporal expression that is
closest to the token in the document, and wherein the at least one
contextual temporal focus of the candidate result generated based
on the definite temporal expression in the content of the document
is identified based on the datetime of tokens in the candidate
result.
6. The method of claim 5, wherein the at least one contextual
temporal focus of the candidate result is identified based on the
datetime of tokens in the candidate result by combining datetimes
of the tokens in the candidate result according to a predetermined
relationship.
7. The method of claim 6, wherein the predetermined relationship is
one of a union of datetimes of tokens in the candidate result or a
minimally overlapping datetime evaluation.
8. The method of claim 4, wherein determining a document relevance
datetime of the document comprises determining the document
relevance datetime in accordance with an established prioritization
of a plurality of types of relevance datetimes.
9. The method of claim 4, wherein the analyzing of the documents is
done as part of an ingestion operation for ingesting the corpus
prior to receipt of the cognitive operation request.
10. The method of claim 1, wherein analyzing the input text to
identify a temporal focus of the input text comprises determining a
current datetime and normalizing definite temporal expressions in
the input text relative to the current datetime.
11. A computer program product comprising a computer readable
storage medium having a computer readable program stored therein,
wherein the computer readable program, when executed on a computing
device, causes the computing device to: receive a cognitive
operation request comprising a portion of input text; analyze the
input text to identify a temporal focus of the input text; process
a corpus of content based on the input portion of text to generate
candidate results; process each candidate result, to identify at
least one contextual temporal focus associated with the candidate
result; compare the at least one contextual temporal focus with the
temporal focus of the input text; generate a measure of temporal
relevance of the candidate result based on results of the
comparison; and perform the cognitive operation based on the
measure of temporal relevance.
12. The computer program product of claim 11, wherein the input
text is a natural language question that has at least one term
corresponding to a temporal focus of the natural language question,
and wherein the candidate results are candidate answers to the
natural language question generated from documents of the
corpus.
13. The computer program product of claim 11, wherein the input
text comprises at least one search term of a search query, and
wherein the at least one search term comprises a term corresponding
to a temporal focus of the search query, and wherein the candidate
results are candidate search results of the search query generated
from documents of the corpus.
14. The computer program product of claim 11, wherein the computer
readable program further causes the computing device to analyze the
documents of the corpus, wherein analyzing the documents of the
corpus comprises, for each document: determining a document
relevance datetime of the document based on at least one of
metadata or content of the document, wherein the document relevance
datetime applies to all content of the document; and associating,
with each definite temporal expression in content of the document,
a datetime with the definite temporal expression, and wherein the
at least one contextual temporal focus of a candidate result
generated based on a definite temporal expression in the content of
the document is identified based on at least one of the document
relevance datetime or the datetime of the definite temporal
expression.
15. The computer program product of claim 14, wherein analyzing the
documents of the corpus further comprises, for each document:
associating with each token in the content of the document, a
datetime based on a document relevance datetime associated with the
document and a datetime associated with a definite temporal
expression that is closest to the token in the document, and
wherein the at least one contextual temporal focus of the candidate
result generated based on the definite temporal expression in the
content of the document is identified based on the datetime of
tokens in the candidate result.
16. The computer program product of claim 15, wherein the at least
one contextual temporal focus of the candidate result is identified
based on the datetime of tokens in the candidate result by
combining datetimes of the tokens in the candidate result according
to a predetermined relationship, wherein the predetermined
relationship is one of a union of datetimes of tokens in the
candidate result or a minimally overlapping datetime
evaluation.
17. The computer program product of claim 14, wherein determining a
document relevance datetime of the document comprises determining
the document relevance datetime in accordance with an established
prioritization of a plurality of types of relevance datetimes.
18. The computer program product of claim 14, wherein the analyzing
of the documents is done as part of an ingestion operation for
ingesting the corpus prior to receipt of the cognitive operation
request.
19. The computer program product of claim 11, wherein the computer
readable program causes the computing device to analyze the input
text to identify a temporal focus of the input text at least by
determining a current datetime and normalizing definite temporal
expressions in the input text relative to the current datetime.
20. An apparatus comprising: a processor; and a memory coupled to
the processor, wherein the memory comprises instructions which,
when executed by the processor, cause the processor to: receive a
cognitive operation request comprising a portion of input text;
analyze the input text to identify a temporal focus of the input
text; process a corpus of content based on the input portion of
text to generate candidate results; process each candidate result,
to identify at least one contextual temporal focus associated with
the candidate result; compare the at least one contextual temporal
focus with the temporal focus of the input text; generate a measure
of temporal relevance of the candidate result based on results of
the comparison; and perform the cognitive operation based on the
measure of temporal relevance.
Description
BACKGROUND
[0002] The present application relates generally to an improved
data processing apparatus and method and more specifically to
mechanisms for evaluating the temporal relevance of answers in a
cognitive system.
[0003] With the increased usage of computing networks, such as the
Internet, humans are currently inundated and overwhelmed with the
amount of information available to them from various structured and
unstructured sources. However, information gaps abound as users try
to piece together what they can find that they believe to be
relevant during searches for information on various subjects. To
assist with such searches, recent research has been directed to
generating Question and Answer (QA) systems which may take an input
question, analyze it, and return results indicative of the most
probable answer to the input question. QA systems provide automated
mechanisms for searching through large sets of sources of content,
e.g., electronic documents, and analyze them with regard to an
input question to determine an answer to the question and a
confidence measure as to how accurate an answer is for answering
the input question.
[0004] Examples, of QA systems are Siri.RTM. from Apple.RTM.,
Cortana.RTM. from Microsoft.RTM., and question answering pipeline
of the IBM Watson.TM. cognitive system available from International
Business Machines (IBM.RTM.) Corporation of Armonk, N.Y. The IBM
Watson.TM. system is an application of advanced natural language
processing, information retrieval, knowledge representation and
reasoning, and machine learning technologies to the field of open
domain question answering. The IBM Watson.TM. system is built on
IBM's DeepQA.TM. technology used for hypothesis generation, massive
evidence gathering, analysis, and scoring. DeepQA.TM. takes an
input question, analyzes it, decomposes the question into
constituent parts, generates one or more hypothesis based on the
decomposed question and results of a primary search of answer
sources, performs hypothesis and evidence scoring based on a
retrieval of evidence from evidence sources, performs synthesis of
the one or more hypothesis, and based on trained models, performs a
final merging and ranking to output an answer to the input question
along with a confidence measure.
SUMMARY
[0005] This Summary is provided to introduce a selection of
concepts in a simplified form that are further described herein in
the Detailed Description. This Summary is not intended to identify
key factors or essential features of the claimed subject matter,
nor is it intended to be used to limit the scope of the claimed
subject matter.
[0006] In one illustrative embodiment, a method is provided, in a
data processing system comprising at least one processor and a
memory, the memory comprising instructions which are executed by
the at least one processor to specifically configure the at least
one processor to implement a cognitive system comprising a temporal
relevance evaluation engine for evaluating a temporal relevance of
a portion of content to a cognitive operation request. The method
comprises receiving, by the cognitive system, a cognitive operation
request comprising a portion of input text and analyzing, by the
temporal relevance evaluation engine, the input text to identify a
temporal focus of the input text. The method further comprises
processing, by the temporal relevance evaluation engine, a corpus
of content based on the input portion of text to generate candidate
results and processing, by the temporal relevance evaluation
engine, each candidate result, to identify at least one contextual
temporal focus associated with the candidate result. The method
also comprises comparing, by the temporal relevance evaluation
engine, the at least one contextual temporal focus with the
temporal focus of the input text and generating, by the temporal
relevance evaluation engine, a measure of temporal relevance of the
candidate result based on results of the comparison. Moreover, the
method comprises performing, by the cognitive system, the cognitive
operation based on the measure of temporal relevance.
[0007] In other illustrative embodiments, a computer program
product comprising a computer useable or readable medium having a
computer readable program is provided. The computer readable
program, when executed on a computing device, causes the computing
device to perform various ones of, and combinations of, the
operations outlined above with regard to the method illustrative
embodiment.
[0008] In yet another illustrative embodiment, a system/apparatus
is provided. The system/apparatus may comprise one or more
processors and a memory coupled to the one or more processors. The
memory may comprise instructions which, when executed by the one or
more processors, cause the one or more processors to perform
various ones of, and combinations of, the operations outlined above
with regard to the method illustrative embodiment.
[0009] These and other features and advantages of the present
invention will be described in, or will become apparent to those of
ordinary skill in the art in view of, the following detailed
description of the example embodiments of the present
invention.
BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS
[0010] The invention, as well as a preferred mode of use and
further objectives and advantages thereof, will best be understood
by reference to the following detailed description of illustrative
embodiments when read in conjunction with the accompanying
drawings, wherein:
[0011] FIG. 1 depicts a schematic diagram of one illustrative
embodiment of a question/answer creation (QA) system in a computer
network;
[0012] FIG. 2 is a block diagram of an example data processing
system in which aspects of the illustrative embodiments are
implemented;
[0013] FIG. 3 illustrates a QA system pipeline for processing an
input question in accordance with one illustrative embodiment;
[0014] FIG. 4 is a flowchart outlining an example operation for
ingesting documents of a corpus and associating datetimes with
tokens in the documents in accordance with one illustrative
embodiment; and
[0015] FIG. 5 is a flowchart outlining an example operation for
evaluating candidate answers to an input question based on a
temporal focus of the input question and contextual temporal foci
of candidate answers in accordance with one illustrative
embodiment.
DETAILED DESCRIPTION
[0016] The illustrative embodiments provide mechanisms for
evaluating the temporal relevance of candidate answers in a
question answering (QA) system. In particular, the illustrative
embodiments provide mechanisms for scoring the temporal relevance
of information provided in an input question to temporal
information present in association with candidate answers in one or
more documents of a corpus of documents, to determine which
candidate answers are more temporally relevant to the input
question than others.
[0017] In question answering systems, to generate answer candidates
for a question, likely answers in the form of words and short
phrases are extracted from a corpus of documents. These candidate
answers are scored and ranked with top ranked answers being
returned to the user as answers to the original question. Many
questions contain time or date information which is crucial to
providing a correct answer. For example, to correctly answer a
question such as "Who was nominated as the Republican candidate for
president in 1976?" answers which talk about both the nominee and
the year 1976 should be ranked above those that talk about the
nominee and the year 2000. More generally, the relationship between
the temporal focus of the question and the temporal focus of the
candidate answer is crucial to ranking.
[0018] For example, consider the following passage: [0019] Ford and
Reagan engaged in a bitter and close fight for the nomination
during the first eight months of 1976, trading victories in a
series of state Republican primaries. As the incumbent, Ford had
courted wavering Republican delegates in key states by inviting
them to the White House, by offering to speak in their states, and
by rewarding delegates with patronage positions. Ford won the
nomination on the first ballot but only by a mere sixty delegate
votes. To determine the temporal focus of the candidate answer
"Ford" as being 1976 requires mechanisms that can process the
entire passage and evaluate the candidate answer based on the
relationship between the temporal focus of the question and the
temporal focus of the candidate answer. The illustrative
embodiments provide such mechanisms for evaluating temporal foci
when generating answers to input questions via a cognitive
system.
[0020] Within the context of the present description, the term
"temporal focus" of a portion of text refers to the definite time
or interval of time, often a date, time, or date/time range, which
is the period for which a claim made by the text naturally applies.
For example, in the text "Bill Clinton was president of the USA in
1995," the temporal focus is "1995."
[0021] The term "datetime" refers to a specification of a definite
period of time, such as a year ("2015"), a day ("2015-09-01"), an
hour ("2015-09-01T12"), a second ("2015-09-01T12:01:30"), or the
like. These may be represented in ISO-8601 standard strings, for
example. Datetimes stand in four important relations to one
another: before (one datetime is before another), after (one
datetime is after another), inclusion (one datetime is within the
range of another datetime), and overlap (one datetime range
overlaps another datetime).
[0022] The term "definite temporal expression" refers to a short
phrase, e.g., "July 4", "Monday", or "yesterday", that refers to a
datetime. A "corpus" or "document corpus" or "corpus of documents"
refers to a set of documents which have been ingested into a
cognitive system, potentially over a protracted period of time.
Typically, these documents will have metadata associated with them
concerning their publication date, creation date, or other temporal
information. The documents will have at least a date and/or time
associated with them at which point the document was ingested into
the cognitive system.
[0023] The term "document" refers to any portion of content which
is stored in an electronic form. A document may range from a few
characters, words, or terms, to sentences, paragraphs, pages,
collections of pages, and so on. A document may comprise textual
and non-textual content including images, video, audio content, or
the like. The document may be stored in any electronic form but in
general will be stored as a portion of data which may have
associated metadata.
[0024] As noted above, one aspect of a candidate answer's
correctness for answering an input question is its temporal
relevance to the focus of the input question, i.e. a match between
the question's temporal focus and the temporal focus of the text
from which the answer is extracted, also referred to herein as the
contextual temporal focus of the candidate answer. In order to
appropriately score candidate answers for their temporal relevance,
each candidate answer extracted from a text has its contextual
temporal focus determined using the mechanisms of the illustrative
embodiments. The determination of a contextual temporal focus can
be quite difficult since, as illustrated by the example given
above, the contextual temporal focus of a candidate answer can be
indicated by textual content which is quite distance from the text
in which the candidate answer is present.
[0025] In accordance with one illustrative embodiment, in order to
determine the contextual temporal focus of a candidate answer, the
mechanisms of the illustrative embodiment first identify an
appropriate document relevance datetime, such as a publication
date/time, ingestion date/time, a datetime associated with a
collection, source, or corpus in which the document is present, or
the like, for the document. This operation may have been done prior
to processing the input question, such as part of an ingestion
operation or may be done as part of the processing of an input
question and may be directed to a document in which the candidate
answer was found, or from which it was extracted. This operation
may comprise analyzing metadata associated with the document to
extract dates/times that are associated with the document and then
select one, if there is more than one, which is most appropriate
for use as a document relevance datetime. It should be appreciated
that if there is more than one date/time associated with the
document, a priority or preference ordering of date/times may be
established for selecting a date/time from those available. For
example, a preference ordering may be established, such as via
configuration information that prioritizes a publication datetime
of the document over a creation datetime of the document, which is
further prioritized over an ingestion datetime of the document.
[0026] The mechanisms also normalize all the definite temporal
expression in the document by associating a datetime with each such
expression. In other words, each temporal expression is associated
with a datetime that is specified in an absolute time value and
uniform format such that datetimes may be accurately compared. This
is significantly different from any known mechanism since known
mechanisms, if they consider temporal aspects of a document at all,
generally attribute all content within a document to the temporal
characteristic of the document itself, e.g., publication date. That
is, known mechanisms at most consider the content of a document to
be co-temporal with the document itself. Moreover, as with the
determination of the document relevance datetime, this operation
may be performed as part of an ingestion operation when ingesting a
document from a corpus, or may be performed as part of the
processing of an input question.
[0027] Furthermore, the mechanisms of this illustrative embodiment
may also associate with each token in the document, e.g., word or
group of alphanumeric characters, one or more temporal foci, making
use of the document relevance datetime and the normalized temporal
expressions associated with the portion of text in which the token
is present or which is closest to the token from the standpoint of
distance measured as a number of tokens (e.g., words), as well as
considering other characteristics of the token and surrounding text
including syntactic structure, matching verb tense between token
and text corresponding to normalized temporal expressions, and the
like. Again, these operations may be performed either at a time of
ingestion of the document or as part of processing an input
question.
[0028] In addition, the mechanisms of the illustrative embodiments
determine one or more temporal foci of the input question. The
identification of the one or more temporal foci may comprise using
the current datetime as the relevant contextual datetime for the
input question. The mechanisms may then identify and normalize all
definite temporal expressions in the question with respect to this
relevant contextual datetime, e.g., if the question ask about "last
year" and the relevant contextual datetime of the question is 2016,
then "last year" would be referring to the year "2015." These
normalized datetimes of the definite temporal expressions are the
temporal foci of the input question. If there are no definite
temporal expressions in the input question, the current datetime
may be selected for present tense questions, otherwise no datetime
is selected. These operations are performed when processing the
input question in response to it being received by the cognitive
system of the illustrative embodiment.
[0029] The mechanisms of the illustrative embodiment may further
process the input question and generate one or more candidate
answers by extracting the candidate answers from the documents of
the corpus. Either previously, through operation of an ingestion
process in which the above operations are performed to associate
datetimes with tokens in the documents, or as part of the
processing of the input question, the tokens that make up the
candidate answers are used to generate one or more temporal foci of
the corresponding candidate answer. Thus, for example, a candidate
answer may have a single word that represents the candidate answer.
That word may, through the operations performed above, have a
temporal focus associated with the candidate answer. If more than
one temporal focus is associated with tokens of the candidate
answer, then the temporal focus of the candidate answer may be
generated based on a predetermined relationship evaluation of the
temporal foci. In one illustrative embodiment, this may be simply a
union of the temporal foci of the various tokens. In other
illustrative embodiments, a more complex relationship evaluation
may be performed on the temporal foci, such as a minimally
overlapping datetime evaluation, and may even associate other
temporal terms that cover a combination of the temporal foci, such
as "before" or "after", e.g., "before 1976" the answer was X or
"after last Monday" the answer is Y.
[0030] Having determined a temporal focus for the input question
and a contextual temporal focus for each of the candidate answers,
the candidate answers are then scored according to the temporal
relevance of the candidate answer with respect to the input
question. For example, in one illustrative embodiment, the
candidate answer may be given a first score, e.g., a "1", if there
is a datetime in the temporal focus or foci of the input question
which overlaps the datetime in the temporal focus of the candidate
answer (contextual temporal focus). Otherwise, if there is no
overlap of this nature, then the candidate answer may be given a
second score, e.g., "0". It should be appreciated that this is only
one simply example. More complex scoring may be used as well based
on how close the temporal foci if the input question are to the
contextual temporal focus of the candidate answer such that a range
of scores between the first and second scores may be assigned to a
candidate answer. For example, a temporal proximity of the temporal
foci of the input question to the contextual temporal focus of the
candidate answer may be evaluated such that candidate answers that
are more remotely proximate to the temporal foci of the input
question are scored lower than those that are more closely
proximate to the temporal foci of the input question. Various other
metrics for scoring candidate answers with regard to temporal
relevance to the input question may be used without departing from
the spirit and scope of the present invention.
[0031] It should also be appreciated that the temporal focus based
scoring of candidate answers may be used as part of a more complex
scoring of candidate answers, such as may be performed by known or
later developed cognitive systems and question answering (QA)
systems. For example, the IBM Watson.TM. cognitive system includes
a QA system which scores candidate answers based on a variety of
factors. The present temporal focus based scoring may be integrated
into a cognitive system and/or QA system, such as IBM Watson.TM. as
an additional factor that is evaluated when scoring candidate
answers. In such a case, various weightings may be attributed to
the temporal focus based on the particular implementation. For
example, in some implementations, the temporal focus may be used as
a basis for essentially "ruling out" certain candidate answers,
e.g., if the candidate answer's contextual temporal focus is not
within the range of the temporal foci of the input question, i.e.
there is no overlap of the temporal foci of the input question with
the contextual temporal focus of the candidate answer, then the
candidate answer may be discarded. In other implementations, the
scoring of the candidate answer on the basis of the contextual
temporal focus of the candidate answer may be added to the overall
scoring of the candidate answer with regard to other factors in
order to generate an overall score for the candidate answer for
purposes of later ranking of candidate answers. This combination of
scoring of various factors may be weighted according to a
predetermined degree of influence of each factor over the
correctness of a candidate answer such that, for example, in some
implementations the contextual temporal focus evaluation may have
greater influence than in other implementations.
[0032] Thus, the illustrative embodiments provide mechanism for
scoring candidate answers based on the temporal relevance of the
candidate answer to temporal foci of an input question. In this
way, candidate answers that are more relevant to the temporal
aspects of the input question may be identified such that the most
relevant candidate answer may be selected as higher ranking or even
final answers for responding to the input question.
[0033] Before beginning the discussion of the various aspects of
the illustrative embodiments in more detail, it should first be
appreciated that throughout this description the term "mechanism"
will be used to refer to elements of the present invention that
perform various operations, functions, and the like. A "mechanism,"
as the term is used herein, may be an implementation of the
functions or aspects of the illustrative embodiments in the form of
an apparatus, a procedure, or a computer program product. In the
case of a procedure, the procedure is implemented by one or more
devices, apparatus, computers, data processing systems, or the
like. In the case of a computer program product, the logic
represented by computer code or instructions embodied in or on the
computer program product is executed by one or more hardware
devices in order to implement the functionality or perform the
operations associated with the specific "mechanism." Thus, the
mechanisms described herein may be implemented as specialized
hardware, software executing on general purpose hardware, software
instructions stored on a medium such that the instructions are
readily executable by specialized or general purpose hardware, a
procedure or method for executing the functions, or a combination
of any of the above.
[0034] The present description and claims may make use of the terms
"a", "at least one of", and "one or more of" with regard to
particular features and elements of the illustrative embodiments.
It should be appreciated that these terms and phrases are intended
to state that there is at least one of the particular feature or
element present in the particular illustrative embodiment, but that
more than one can also be present. That is, these terms/phrases are
not intended to limit the description or claims to a single
feature/element being present or require that a plurality of such
features/elements be present. To the contrary, these terms/phrases
only require at least a single feature/element with the possibility
of a plurality of such features/elements being within the scope of
the description and claims.
[0035] Moreover, it should be appreciated that the use of the term
"engine," if used herein with regard to describing embodiments and
features of the invention, is not intended to be limiting of any
particular implementation for accomplishing and/or performing the
actions, steps, processes, etc., attributable to and/or performed
by the engine. An engine may be, but is not limited to, software,
hardware and/or firmware or any combination thereof that performs
the specified functions including, but not limited to, any use of a
general and/or specialized processor in combination with
appropriate software loaded or stored in a machine readable memory
and executed by the processor. Further, any name associated with a
particular engine is, unless otherwise specified, for purposes of
convenience of reference and not intended to be limiting to a
specific implementation. Additionally, any functionality attributed
to an engine may be equally performed by multiple engines,
incorporated into and/or combined with the functionality of another
engine of the same or different type, or distributed across one or
more engines of various configurations.
[0036] In addition, it should be appreciated that the following
description uses a plurality of various examples for various
elements of the illustrative embodiments to further illustrate
example implementations of the illustrative embodiments and to aid
in the understanding of the mechanisms of the illustrative
embodiments. These examples intended to be non-limiting and are not
exhaustive of the various possibilities for implementing the
mechanisms of the illustrative embodiments. It will be apparent to
those of ordinary skill in the art in view of the present
description that there are many other alternative implementations
for these various elements that may be utilized in addition to, or
in replacement of, the examples provided herein without departing
from the spirit and scope of the present invention.
[0037] The present invention may be a system, a method, and/or a
computer program product. The computer program product may include
a computer readable storage medium (or media) having computer
readable program instructions thereon for causing a processor to
carry out aspects of the present invention.
[0038] The computer readable storage medium can be a tangible
device that can retain and store instructions for use by an
instruction execution device. The computer readable storage medium
may be, for example, but is not limited to, an electronic storage
device, a magnetic storage device, an optical storage device, an
electromagnetic storage device, a semiconductor storage device, or
any suitable combination of the foregoing. A non-exhaustive list of
more specific examples of the computer readable storage medium
includes the following: a portable computer diskette, a hard disk,
a random access memory (RAM), a read-only memory (ROM), an erasable
programmable read-only memory (EPROM or Flash memory), a static
random access memory (SRAM), a portable compact disc read-only
memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a
floppy disk, a mechanically encoded device such as punch-cards or
raised structures in a groove having instructions recorded thereon,
and any suitable combination of the foregoing. A computer readable
storage medium, as used herein, is not to be construed as being
transitory signals per se, such as radio waves or other freely
propagating electromagnetic waves, electromagnetic waves
propagating through a waveguide or other transmission media (e.g.,
light pulses passing through a fiber-optic cable), or electrical
signals transmitted through a wire.
[0039] Computer readable program instructions described herein can
be downloaded to respective computing/processing devices from a
computer readable storage medium or to an external computer or
external storage device via a network, for example, the Internet, a
local area network, a wide area network and/or a wireless network.
The network may comprise copper transmission cables, optical
transmission fibers, wireless transmission, routers, firewalls,
switches, gateway computers and/or edge servers. A network adapter
card or network interface in each computing/processing device
receives computer readable program instructions from the network
and forwards the computer readable program instructions for storage
in a computer readable storage medium within the respective
computing/processing device.
[0040] Computer readable program instructions for carrying out
operations of the present invention may be assembler instructions,
instruction-set-architecture (ISA) instructions, machine
instructions, machine dependent instructions, microcode, firmware
instructions, state-setting data, or either source code or object
code written in any combination of one or more programming
languages, including an object oriented programming language such
as Java, Smalltalk, C++ or the like, and conventional procedural
programming languages, such as the "C" programming language or
similar programming languages. The computer readable program
instructions may execute entirely on the user's computer, partly on
the user's computer, as a stand-alone software package, partly on
the user's computer and partly on a remote computer or entirely on
the remote computer or server. In the latter scenario, the remote
computer may be connected to the user's computer through any type
of network, including a local area network (LAN) or a wide area
network (WAN), or the connection may be made to an external
computer (for example, through the Internet using an Internet
Service Provider). In some embodiments, electronic circuitry
including, for example, programmable logic circuitry,
field-programmable gate arrays (FPGA), or programmable logic arrays
(PLA) may execute the computer readable program instructions by
utilizing state information of the computer readable program
instructions to personalize the electronic circuitry, in order to
perform aspects of the present invention.
[0041] Aspects of the present invention are described herein with
reference to flowchart illustrations and/or block diagrams of
methods, apparatus (systems), and computer program products
according to embodiments of the invention. It will be understood
that each block of the flowchart illustrations and/or block
diagrams, and combinations of blocks in the flowchart illustrations
and/or block diagrams, can be implemented by computer readable
program instructions.
[0042] These computer readable program instructions may be provided
to a processor of a general purpose computer, special purpose
computer, or other programmable data processing apparatus to
produce a machine, such that the instructions, which execute via
the processor of the computer or other programmable data processing
apparatus, create means for implementing the functions/acts
specified in the flowchart and/or block diagram block or blocks.
These computer readable program instructions may also be stored in
a computer readable storage medium that can direct a computer, a
programmable data processing apparatus, and/or other devices to
function in a particular manner, such that the computer readable
storage medium having instructions stored therein comprises an
article of manufacture including instructions which implement
aspects of the function/act specified in the flowchart and/or block
diagram block or blocks.
[0043] The computer readable program instructions may also be
loaded onto a computer, other programmable data processing
apparatus, or other device to cause a series of operational steps
to be performed on the computer, other programmable apparatus or
other device to produce a computer implemented process, such that
the instructions which execute on the computer, other programmable
apparatus, or other device implement the functions/acts specified
in the flowchart and/or block diagram block or blocks.
[0044] The flowchart and block diagrams in the Figures illustrate
the architecture, functionality, and operation of possible
implementations of systems, methods, and computer program products
according to various embodiments of the present invention. In this
regard, each block in the flowchart or block diagrams may represent
a module, segment, or portion of instructions, which comprises one
or more executable instructions for implementing the specified
logical function(s). In some alternative implementations, the
functions noted in the block may occur out of the order noted in
the figures. For example, two blocks shown in succession may, in
fact, be executed substantially concurrently, or the blocks may
sometimes be executed in the reverse order, depending upon the
functionality involved. It will also be noted that each block of
the block diagrams and/or flowchart illustration, and combinations
of blocks in the block diagrams and/or flowchart illustration, can
be implemented by special purpose hardware-based systems that
perform the specified functions or acts or carry out combinations
of special purpose hardware and computer instructions.
[0045] The illustrative embodiments may be utilized in many
different types of data processing environments. In order to
provide a context for the description of the specific elements and
functionality of the illustrative embodiments, FIGS. 1-3 are
provided hereafter as example environments in which aspects of the
illustrative embodiments may be implemented. It should be
appreciated that FIGS. 1-3 are only examples and are not intended
to assert or imply any limitation with regard to the environments
in which aspects or embodiments of the present invention may be
implemented. Many modifications to the depicted environments may be
made without departing from the spirit and scope of the present
invention.
[0046] FIGS. 1-3 are directed to describing an example cognitive
system implementing a Question Answering (QA) pipeline (also
referred to as a Question/Answer pipeline or Question and Answer
pipeline), methodology, and computer program product with which the
mechanisms of the illustrative embodiments are implemented. The QA
pipeline is part of a QA system that may be implemented in the
cognitive system. The cognitive system, while shown as having a
single QA pipeline, may in fact have multiple QA pipelines. Each QA
pipeline may be separately trained for answer input questions of a
different domains or be configured to perform the same or different
analysis on input questions, depending on the desired
implementation. For example, in some cases, a first QA pipeline may
be trained to operate on input questions in a financial domain
while another QA pipeline may be trained to answer input questions
in a medical diagnostics domain. Moreover, each QA pipeline may
have their own associated corpus or corpora that they ingest and
operate on, e.g., one corpus for financial domain documents and
another corpus for medical diagnostics domain related documents in
the above examples. In some cases, the QA pipelines may each
operate on the same domain of input questions but may have
different configurations, e.g., different annotators or differently
trained annotators, such that different analysis and potential
answers are generated. The QA system may provide additional logic
for routing input questions to the appropriate QA pipeline, such as
based on a determined domain of the input question, combining and
evaluating final answers generated by multiple QA pipelines, and
other control and interaction logic that facilitates the
utilization of multiple QA pipelines.
[0047] As will be discussed in greater detail hereafter, the
illustrative embodiments are integrated in, augment, and extend the
functionality of these QA mechanisms of the cognitive system with
regard to evaluating the temporal relevance of candidate answers to
the temporal foci of an input question. These mechanisms extend the
functionality by providing logic for identifying tokens in
documents and associating with these tokens one or more temporal
foci. The mechanisms further extend the functionality by
identifying one or more temporal foci of the input question,
associating with candidate answers a contextual temporal focus
based on the temporal foci of the tokens associated with the
candidate answer, and then score the candidate answers based on the
relevance of the contextual temporal focus of the candidate answers
to the one or more temporal foci of the input question. In this
way, more temporally relevant candidate answers are ranked higher
than less or non-temporally relevant candidate answers. Thus, more
improved answer results are generated.
[0048] Since the present invention extends the functionality of a
QA system, it is important to first have an understanding of how
question and answer creation in a cognitive system implementing a
QA pipeline is implemented before describing how the mechanisms of
the illustrative embodiments are integrated in and augment such QA
mechanisms. It should be appreciated that the QA mechanisms
described in FIGS. 1-3 are only examples and are not intended to
state or imply any limitation with regard to the type of QA
mechanisms with which the illustrative embodiments are implemented.
Many modifications to the example cognitive system shown in FIGS.
1-3 may be implemented in various embodiments of the present
invention without departing from the spirit and scope of the
present invention.
[0049] As an overview, a cognitive system is a specialized computer
system, or set of computer systems, configured with hardware and/or
software logic (in combination with hardware logic upon which the
software executes) to emulate human cognitive functions. These
cognitive systems apply human-like characteristics to conveying and
manipulating ideas which, when combined with the inherent strengths
of digital computing, can solve problems with high accuracy and
resilience on a large scale. A cognitive system performs one or
more computer-implemented cognitive operations that approximate a
human thought process as well as enable people and machines to
interact in a more natural manner so as to extend and magnify human
expertise and cognition. A cognitive system comprises artificial
intelligence logic, such as natural language processing (NLP) based
logic, for example, and machine learning logic, which may be
provided as specialized hardware, software executed on hardware, or
any combination of specialized hardware and software executed on
hardware. The logic of the cognitive system implements the
cognitive operation(s), examples of which include, but are not
limited to, question answering, identification of related concepts
within different portions of content in a corpus, intelligent
search algorithms, such as Internet web page searches, for example,
medical diagnostic and treatment recommendations, and other types
of recommendation generation, e.g., items of interest to a
particular user, potential new contact recommendations, or the
like.
[0050] IBM Watson.TM. is an example of one such cognitive system
which can process human readable language and identify inferences
between text passages with human-like high accuracy at speeds far
faster than human beings and on a larger scale. In general, such
cognitive systems are able to perform the following functions:
[0051] Navigate the complexities of human language and
understanding [0052] Ingest and process vast amounts of structured
and unstructured data [0053] Generate and evaluate hypothesis
[0054] Weigh and evaluate responses that are based only on relevant
evidence [0055] Provide situation-specific advice, insights, and
guidance [0056] Improve knowledge and learn with each iteration and
interaction through machine learning processes [0057] Enable
decision making at the point of impact (contextual guidance) [0058]
Scale in proportion to the task [0059] Extend and magnify human
expertise and cognition [0060] Identify resonating, human-like
attributes and traits from natural language [0061] Deduce various
language specific or agnostic attributes from natural language
[0062] High degree of relevant recollection from data points
(images, text, voice) (memorization and recall) [0063] Predict and
sense with situational awareness that mimic human cognition based
on experiences [0064] Answer questions based on natural language
and specific evidence
[0065] In one aspect, cognitive systems provide mechanisms for
answering questions posed to these cognitive systems using a
Question Answering pipeline or system (QA system). The QA pipeline
or system is an artificial intelligence application executing on
data processing hardware that answers questions pertaining to a
given subject-matter domain presented in natural language. The QA
pipeline receives inputs from various sources including input over
a network, a corpus of electronic documents or other data, data
from a content creator, information from one or more content users,
and other such inputs from other possible sources of input. Data
storage devices store the corpus of data. A content creator creates
content in a document for use as part of a corpus of data with the
QA pipeline. The document may include any file, text, article, or
source of data for use in the QA system. For example, a QA pipeline
accesses a body of knowledge about the domain, or subject matter
area, e.g., financial domain, medical domain, legal domain, etc.,
where the body of knowledge (knowledgebase) can be organized in a
variety of configurations, e.g., a structured repository of
domain-specific information, such as ontologies, or unstructured
data related to the domain, or a collection of natural language
documents about the domain.
[0066] Content users input questions to cognitive system which
implements the QA pipeline. The QA pipeline then answers the input
questions using the content in the corpus of data by evaluating
documents, sections of documents, portions of data in the corpus,
or the like. When a process evaluates a given section of a document
for semantic content, the process can use a variety of conventions
to query such document from the QA pipeline, e.g., sending the
query to the QA pipeline as a well-formed question which is then
interpreted by the QA pipeline and a response is provided
containing one or more answers to the question. Semantic content is
content based on the relation between signifiers, such as words,
phrases, signs, and symbols, and what they stand for, their
denotation, or connotation. In other words, semantic content is
content that interprets an expression, such as by using Natural
Language Processing.
[0067] As will be described in greater detail hereafter, the QA
pipeline receives an input question, parses the question to extract
the major features of the question, uses the extracted features to
formulate queries, and then applies those queries to the corpus of
data. Based on the application of the queries to the corpus of
data, the QA pipeline generates a set of hypotheses, or candidate
answers to the input question, by looking across the corpus of data
for portions of the corpus of data that have some potential for
containing a valuable response to the input question. The QA
pipeline then performs deep analysis on the language of the input
question and the language used in each of the portions of the
corpus of data found during the application of the queries using a
variety of reasoning algorithms. There may be hundreds or even
thousands of reasoning algorithms applied, each of which performs
different analysis, e.g., comparisons, natural language analysis,
lexical analysis, or the like, and generates a score. For example,
some reasoning algorithms may look at the matching of terms and
synonyms within the language of the input question and the found
portions of the corpus of data. Other reasoning algorithms may look
at temporal or spatial features in the language, while others may
evaluate the source of the portion of the corpus of data and
evaluate its veracity.
[0068] The scores obtained from the various reasoning algorithms
indicate the extent to which the potential response is inferred by
the input question based on the specific area of focus of that
reasoning algorithm. Each resulting score is then weighted against
a statistical model. The statistical model captures how well the
reasoning algorithm performed at establishing the inference between
two similar passages for a particular domain during the training
period of the QA pipeline. The statistical model is used to
summarize a level of confidence that the QA pipeline has regarding
the evidence that the potential response, i.e. candidate answer, is
inferred by the question. This process is repeated for each of the
candidate answers until the QA pipeline identifies candidate
answers that surface as being significantly stronger than others
and thus, generates a final answer, or ranked set of answers, for
the input question.
[0069] As mentioned above, QA pipeline and mechanisms operate by
accessing information from a corpus of data or information (also
referred to as a corpus of content), analyzing it, and then
generating answer results based on the analysis of this data.
Accessing information from a corpus of data typically includes: a
database query that answers questions about what is in a collection
of structured records, and a search that delivers a collection of
document links in response to a query against a collection of
unstructured data (text, markup language, etc.). Conventional
question answering systems are capable of generating answers based
on the corpus of data and the input question, verifying answers to
a collection of questions for the corpus of data, correcting errors
in digital text using a corpus of data, and selecting answers to
questions from a pool of potential answers, i.e. candidate
answers.
[0070] Content creators, such as article authors, electronic
document creators, web page authors, document database creators,
and the like, determine use cases for products, solutions, and
services described in such content before writing their content.
Consequently, the content creators know what questions the content
is intended to answer in a particular topic addressed by the
content. Categorizing the questions, such as in terms of roles,
type of information, tasks, or the like, associated with the
question, in each document of a corpus of data allows the QA
pipeline to more quickly and efficiently identify documents
containing content related to a specific query. The content may
also answer other questions that the content creator did not
contemplate that may be useful to content users. The questions and
answers may be verified by the content creator to be contained in
the content for a given document. These capabilities contribute to
improved accuracy, system performance, machine learning, and
confidence of the QA pipeline. Content creators, automated tools,
or the like, annotate or otherwise generate metadata for providing
information useable by the QA pipeline to identify these question
and answer attributes of the content.
[0071] Operating on such content, the QA pipeline generates answers
for input questions using a plurality of intensive analysis
mechanisms which evaluate the content to identify the most probable
answers, i.e. candidate answers, for the input question. The most
probable answers are output as a ranked listing of candidate
answers ranked according to their relative scores or confidence
measures calculated during evaluation of the candidate answers, as
a single final answer having a highest ranking score or confidence
measure, or which is a best match to the input question, or a
combination of ranked listing and final answer.
[0072] FIG. 1 depicts a schematic diagram of one illustrative
embodiment of a cognitive system 100 implementing a question
answering (QA) pipeline 108 in a computer network 102. One example
of a question/answer generation operation which may be used in
conjunction with the principles described herein is described in
U.S. Patent Application Publication No. 2011/0125734, which is
herein incorporated by reference in its entirety. The cognitive
system 100 is implemented on one or more computing devices 104
(comprising one or more processors and one or more memories, and
potentially any other computing device elements generally known in
the art including buses, storage devices, communication interfaces,
and the like) connected to the computer network 102. The network
102 includes multiple computing devices 104 in communication with
each other and with other devices or components via one or more
wired and/or wireless data communication links, where each
communication link comprises one or more of wires, routers,
switches, transmitters, receivers, or the like. The cognitive
system 100 and network 102 enables question/answer (QA) generation
functionality for one or more cognitive system users via their
respective computing devices 110-112. Other embodiments of the
cognitive system 100 may be used with components, systems,
sub-systems, and/or devices other than those that are depicted
herein.
[0073] The cognitive system 100 is configured to implement a QA
pipeline 108 that receive inputs from various sources. For example,
the cognitive system 100 receives input from the network 102, a
corpus of electronic documents 106, cognitive system users, and/or
other data and other possible sources of input. In one embodiment,
some or all of the inputs to the cognitive system 100 are routed
through the network 102. The various computing devices 104 on the
network 102 include access points for content creators and QA
system users. Some of the computing devices 104 include devices for
a database storing the corpus of data 106 (which is shown as a
separate entity in FIG. 1 for illustrative purposes only). Portions
of the corpus of data 106 may also be provided on one or more other
network attached storage devices, in one or more databases, or
other computing devices not explicitly shown in FIG. 1. The network
102 includes local network connections and remote connections in
various embodiments, such that the cognitive system 100 may operate
in environments of any size, including local and global, e.g., the
Internet.
[0074] In one embodiment, the content creator creates content in a
document of the corpus of data 106 for use as part of a corpus of
data with the cognitive system 100. The document includes any file,
text, article, or source of data for use in the cognitive system
100. QA system users access the cognitive system 100 via a network
connection or an Internet connection to the network 102, and input
questions to the cognitive system 100 that are answered by the
content in the corpus of data 106. In one embodiment, the questions
are formed using natural language. The cognitive system 100 parses
and interprets the question via a QA pipeline 108, and provides a
response to the cognitive system user, e.g., cognitive system user
110, containing one or more answers to the question. In some
embodiments, the cognitive system 100 provides a response to users
in a ranked list of candidate answers while in other illustrative
embodiments, the cognitive system 100 provides a single final
answer or a combination of a final answer and ranked listing of
other candidate answers.
[0075] The cognitive system 100 implements the QA pipeline 108
which comprises a plurality of stages for processing an input
question and the corpus of data 106. The QA pipeline 108 generates
answers for the input question based on the processing of the input
question and the corpus of data 106. The QA pipeline 108 will be
described in greater detail hereafter with regard to FIG. 3.
[0076] In some illustrative embodiments, the cognitive system 100
may be the IBM Watson.TM. cognitive system available from
International Business Machines Corporation of Armonk, N.Y., which
is augmented with the mechanisms of the illustrative embodiments
described hereafter. As outlined previously, a QA pipeline of the
IBM Watson.TM. cognitive system receives an input question which it
then parses to extract the major features of the question, which in
turn are then used to formulate queries that are applied to the
corpus of data. Based on the application of the queries to the
corpus of data, a set of hypotheses, or candidate answers to the
input question, are generated by looking across the corpus of data
for portions of the corpus of data that have some potential for
containing a valuable response to the input question. The QA
pipeline of the IBM Watson.TM. cognitive system then performs deep
analysis on the language of the input question and the language
used in each of the portions of the corpus of data found during the
application of the queries using a variety of reasoning algorithms.
The scores obtained from the various reasoning algorithms are then
weighted against a statistical model that summarizes a level of
confidence that the QA pipeline of the IBM Watson.TM. cognitive
system has regarding the evidence that the potential response, i.e.
candidate answer, is inferred by the question. This process is be
repeated for each of the candidate answers to generate ranked
listing of candidate answers which may then be presented to the
user that submitted the input question, or from which a final
answer is selected and presented to the user. More information
about the QA pipeline of the IBM Watson.TM. cognitive system may be
obtained, for example, from the IBM Corporation website, IBM
Redbooks, and the like. For example, information about the QA
pipeline of the IBM Watson.TM. cognitive system can be found in
Yuan et al., "Watson.TM. and Healthcare," IBM developerWorks, 2011
and "The Era of Cognitive Systems: An Inside Look at IBM Watson.TM.
and How it Works" by Rob High, IBM Redbooks, 2012.
[0077] As shown in FIG. 1, the cognitive system 100 is further
augmented, in accordance with the mechanisms of the illustrative
embodiments, to include logic implemented in specialized hardware,
software executed on hardware, or any combination of specialized
hardware and software executed on hardware, for implementing a
temporal relevance evaluation engine 120. The temporal relevance
evaluation engine 120 provides logic that may be utilized by the
cognitive system 100 during ingestion of the documents of the
corpus 130 and/or during runtime evaluation of input questions from
client computing devices 110, 112, to evaluate the temporal
characteristics of portions of content of the documents and
associated with these portions of content metadata identifying a
temporal focus or foci of the portion of content. For example, the
documents may be analyzed to identify a document relevance datetime
which is then used as a basis for associating a datetime with each
of the definite temporal expressions in the content of the
document, which is then in turn used, along with the temporal
relevance datetime of the document, as a basis for associating with
each token in the document a closest temporal focus or foci. Thus,
each token of each document may have an associated temporal focus
or foci. In addition, the temporal relevance evaluation engine 120
provides logic that analyzes an input portion of text, e.g., an
input question, input search query, or the like, and identifies one
or more temporal foci of the input portion of text, which for
purposes of the present description will be assumed to be an input
natural language question. The QA system pipeline 108 of the
cognitive system 100 may then process the input question to
generate candidate answers based on the corpus 130 and these
candidate answers may be evaluated to identify a contextual
temporal focus for each of the candidate answers. The contextual
temporal focus is determined based on the temporal focus or foci of
the terms in the candidate answer. The contextual temporal focus of
the candidate answer is then compared to the temporal focus or foci
of the input question and the candidate answer is scored based on
the results of the comparison.
[0078] In order to illustrate the operation of an illustrative
embodiment of the present invention, reference will be repeatedly
made hereafter to an example scenario in which the input question
received by the cognitive system 100 and processed by the QA system
pipeline 108 is "Who was nominated as the Republican candidate for
president in 1976?" In addition, the example portion of content
from a document in the corpus 130 that will be used for purposes of
this running example, will be the passage previously mentioned
above, which will be referred to as passage P in document D: [0079]
Ford and Reagan engaged in a bitter and close fight for the
nomination during the first eight months of 1976, trading victories
in a series of state Republican primaries. As the incumbent, Ford
had courted wavering Republican delegates in key states by inviting
them to the White House, by offering to speak in their states, and
by rewarding delegates with patronage positions. Ford won the
nomination on the first ballot but only by a mere sixty delegate
votes.
[0080] With this example in mind, as discussed previously, a set of
operations are performed to associate with each of the tokens of
documents in the corpus 130, a corresponding temporal focus or foci
that can later be used to determine a contextual temporal focus of
candidate answers that comprise those tokens. In order to generate
such a temporal focus or foci for each of the tokens, the context
of the tokens within the document must be evaluated and the
temporal characteristics of the context must be determined. This
can be done as part of an ingestion operation, such as when the
corpus 130 is ingested by the cognitive system 100 for use by the
QA system pipeline 108, where ingestion is a process of analyzing
natural language content and generating an in-memory representation
of that content. These operations can also be done on an as needed
based, such as during runtime processing of an input question,
e.g., in response to a candidate answer being identified in the
document, the document may then be analyzed in the manner described
herein to identify temporal characteristics and temporal foci
associated with the candidate answer. It should be appreciated that
for ease of explanation, the example above, and the following
description will be provided with regard to a single document,
however this process may be repeated for each document in the
corpus 130 or for each document with which a candidate answer is
associated.
[0081] The document temporal expression datetime normalization
logic 122 performs operations for identifying an appropriate
document relevance datetime for the document and normalizing all
definite temporal expressions in the document by associating a
datetime with each such definite temporal expression. In one
illustrative embodiment, the document temporal expression datetime
normalization logic 122 first identifies an appropriate document
relevance datetime, such as a publication date/time, ingestion
date/time, a datetime associated with a collection, source, or
corpus 130 in which the document is present, or the like, for the
document. This operation may comprise analyzing metadata associated
with the document to extract dates/times that are associated with
the document and then select one, if there is more than one, which
is most appropriate for use as a document relevance datetime. If
there is more than one date/time associated with the document, the
document temporal expression datetime normalization logic 122 may
have been configured with a priority or preference ordering of
date/times to be used for selecting a date/time from those
available. For example, a preference ordering may be established by
configuring the document temporal expression datetime normalization
logic 122 to prioritize a publication datetime of the document over
a creation datetime of the document, which is further prioritized
over an ingestion datetime of the document.
[0082] Thus, using the above example, passage P may be provided in
document D which has a publication datetime of "1985," a creation
datetime of "1983", and an ingestion datetime of "2016." This
information may be present in metadata of the document D which may
be analyzed by the document temporal expression datetime
normalization logic 122 which then, in accordance with its
configured prioritization of datetimes, selects the datetime of
"1985" as the document relevance datetime for document D.
[0083] The document temporal expression datetime normalization
logic 122 also normalizes all the definite temporal expression in
the document D by associating a datetime with each such expression.
This operation first requires the identification of definite
temporal expressions within the document D. Such operations may be
performed by performing a matching operation between content of the
document D and a predetermined set of definite temporal expressions
that are recognized by the document temporal expression datetime
normalization logic 122, which may be specified as one or more
tokens, words, phrases, or the like. Thus, for example, in the
passage P above, the definite temporal expression that may be found
is "first eight months of 1976."
[0084] The normalization performed by the document temporal
expression datetime normalization logic 122 may comprise analyzing
the found definite temporal expressions to determine if a
particular datetime is expressly stated in the expression or if the
definite temporal expression needs to be evaluated relative to the
document relevance date. For example, if the definite temporal
expression mentions a specific date and/or time, e.g., "1976" or
"May 3, 2014", or "01:37 pm on Monday, Apr. 14, 1999", then a
relative evaluation is not necessary and the specific date/time may
be adopted as the datetime for the definite temporal expression.
However, if the definite temporal expression is of the type "last
Monday" or "next week", then those types of expressions indicate a
relative measure that is relative to a datetime associated with the
document itself, i.e. the document relevance datetime. Thus, in the
example passage P above, the definite temporal expression "first
eight months of 1976" mentions a specific date and time range, i.e.
"1976" and "first eight months." The document temporal expression
datetime normalization logic 122 may convert this definite temporal
expression to a datetime of 1976 and/or a range of 01/01/76 to
08/31/76, depending on the particular desired implementation.
[0085] Furthermore, the document token temporal foci logic 124 also
associates with each token in the document, e.g., word or group of
alphanumeric characters, one or more temporal foci, making use of
the document relevance datetime and the normalized temporal
expressions associated with the portion of document D and/or
portion of text, e.g., passage P, in which the token is present or
which is closest to the token from the standpoint of distance
measured as a number of tokens (e.g., words), as well as
considering other characteristics of the token and surrounding text
including syntactic structure, matching verb tense between token
and text corresponding to normalized temporal expressions, and the
like. Thus, for example, in the passage P above, each of the words,
or tokens, in the passage P may have associated with it the
datetime of "1976" since that is the datetime associated with the
closest definite temporal expression in the passage P.
[0086] As mentioned previously, in some illustrative embodiments,
the analysis performed by the logic 122-124 may be performed as
part of an ingestion operation in which documents of the corpus 130
are ingested for use by the cognitive system 100 when processing an
input portion of text, e.g., input question or search query.
Alternatively, the operation of logic 122-124 may be performed
during runtime processing of the input portion of text.
[0087] In addition, the input question temporal foci evaluation
logic 126 determines one or more temporal foci of the input portion
of text, e.g., input natural language question. The identification
of the one or more temporal foci may comprise using the current
datetime as the relevant contextual datetime for the input
question. The mechanisms may then identify and normalize all
definite temporal expressions in the question with respect to this
relevant contextual datetime, e.g., if the question ask about "last
year" and the relevant contextual datetime of the question is 2016,
then "last year" would be referring to the year "2015." The
normalization of the definite temporal expressions may be performed
in a similar manner as described above with regard to normalization
performed by the document temporal expression datetime
normalization logic 122, e.g., determining if the definite temporal
expression specifies a particular datetime and if not, determining
if the definite temporal expression specifies a relative datetime
token, word, phrase, or the like. These normalized datetimes of the
definite temporal expressions are the temporal foci of the input
question. If there are no definite temporal expressions in the
input question, the current datetime may be selected for present
tense questions, otherwise no datetime is selected. These
operations are performed when processing the input question in
response to it being received by the cognitive system 100 of the
illustrative embodiment. Thus, for example, in the example above,
the input question "Who was nominated as the Republican candidate
for president in 1976?" the datetime associated with the input
question is "1976" as it is expressly stated within the input
question itself. A prioritized association of datetime with the
input question may be utilized, such as determining if there is any
datetime specified in the input question itself, determining if
there is a relative datetime specified which can be evaluated
against another datetime specified in the input question or against
the current time, e.g., "first eight months of 1976" or "last
week", using the current datetime for present tense questions,
etc., as noted above.
[0088] The QA system pipeline 108 of the cognitive system 100 may
further process the input question and generate one or more
candidate answers by extracting the candidate answers from the
documents of the corpus 130. The candidate answers may be further
evaluated by the candidate answer temporal foci logic 128 and
candidate answer temporal relevance scoring logic 129 working in
conjunction with the QA system pipeline 108.
[0089] Either previously, through operation of an ingestion process
in which the above operations are performed to associate datetimes
with tokens in the documents, or as part of the processing of the
input question, the tokens that make up the candidate answers are
used to generate one or more temporal foci of the corresponding
candidate answer. That is, the candidate answer temporal foci logic
128 takes the candidate answers generated by the QA system pipeline
108 and identifies the datetimes associated with the tokens that
make up the candidate answer. Thus, for example, a candidate answer
may have a single word that represents the candidate answer, e.g.,
in the passage P of the example above, two candidate answers may be
generated such as "Ford" and "Reagan" and the datetime "1976" may
be associated with both candidate answer tokens from the operation
of the document token temporal foci logic 124. The word or words of
the candidate answer may, through the operations performed above,
be used to generate a temporal focus associated with the candidate
answer. If more than one temporal foci are associated with tokens
of the candidate answer, then the temporal focus of the candidate
answer may be generated by the candidate answer temporal foci logic
128 based on a predetermined relationship evaluation of the
temporal foci. In one illustrative embodiment, this may be simply a
union of the temporal foci of the various tokens. In other
illustrative embodiments, a more complex relationship evaluation
may be performed on the temporal foci, such as a minimally
overlapping datetime evaluation, and may even associate other
temporal terms that cover a combination of the temporal foci, such
as "before" or "after".
[0090] Having determined, by the input question temporal foci
evaluation logic 126, a temporal focus for the input question and a
contextual temporal focus for each of the candidate answers by the
candidate answer temporal foci logic 128, the candidate answers are
then scored by the candidate temporal relevance scoring logic 129
according to the temporal relevance of the candidate answer with
respect to the input question. As noted above, in one illustrative
embodiment, the candidate answer may be given a first score, e.g.,
a "1", if there is a datetime in the temporal focus or foci of the
input question which overlaps the datetime in the temporal focus of
the candidate answer (contextual temporal focus). Otherwise, if
there is no overlap of this nature, then the candidate answer may
be given a second score, e.g., "0". In other illustrative
embodiments, a more complex scoring may be used by the candidate
answer temporal relevance scoring logic 129 which is based on how
close the temporal foci in the input question are to the contextual
temporal focus of the candidate answer, such that a range of scores
between the first and second scores may be assigned to a candidate
answer. Various other metrics for scoring candidate answers with
regard to temporal relevance to the input question may be used
without departing from the spirit and scope of the present
invention.
[0091] It should also be appreciated that the temporal focus based
scoring of candidate answers, performed by the candidate answer
temporal relevance scoring logic 129, may be used as part of a more
complex scoring of candidate answers by the logic of the QA system
pipeline 108 and/or cognitive system 100. For example, the temporal
focus based scoring of the candidate answer temporal relevance
scoring logic 129 may be integrated into the cognitive system 100
and/or QA system pipeline 108 as an additional factor that is
evaluated when scoring candidate answers. In such a case, various
weightings may be attributed to the temporal focus based on the
particular implementation. For example, in some implementations,
the temporal focus may be used as a basis for essentially "ruling
out" certain candidate answers. In other implementations, the
scoring of the candidate answer on the basis of the contextual
temporal focus of the candidate answer may be added to the overall
scoring of the candidate answer with regard to other factors in
order to generate an overall score for the candidate answer for
purposes of later ranking of candidate answers. This combination of
scoring of various factors may be weighted according to a
predetermined degree of influence of each factor over the
correctness of a candidate answer such that, for example, in some
implementations the contextual temporal focus evaluation may have
greater influence than in other implementations.
[0092] FIG. 2 is a block diagram of an example data processing
system in which aspects of the illustrative embodiments are
implemented. Data processing system 200 is an example of a
computer, such as server 104 or client 110 in FIG. 1, in which
computer usable code or instructions implementing the processes for
illustrative embodiments of the present invention are located. In
one illustrative embodiment, FIG. 2 represents a server computing
device, such as a server 104, which, which implements a cognitive
system 100 and QA system pipeline 108 augmented to include the
additional mechanisms of the illustrative embodiments described
hereafter.
[0093] In the depicted example, data processing system 200 employs
a hub architecture including north bridge and memory controller hub
(NB/MCH) 202 and south bridge and input/output (I/O) controller hub
(SB/ICH) 204. Processing unit 206, main memory 208, and graphics
processor 210 are connected to NB/MCH 202. Graphics processor 210
is connected to NB/MCH 202 through an accelerated graphics port
(AGP).
[0094] In the depicted example, local area network (LAN) adapter
212 connects to SB/ICH 204. Audio adapter 216, keyboard and mouse
adapter 220, modem 222, read only memory (ROM) 224, hard disk drive
(HDD) 226, CD-ROM drive 230, universal serial bus (USB) ports and
other communication ports 232, and PCI/PCIe devices 234 connect to
SB/ICH 204 through bus 238 and bus 240. PCI/PCIe devices may
include, for example, Ethernet adapters, add-in cards, and PC cards
for notebook computers. PCI uses a card bus controller, while PCIe
does not. ROM 224 may be, for example, a flash basic input/output
system (BIOS).
[0095] HDD 226 and CD-ROM drive 230 connect to SB/ICH 204 through
bus 240. HDD 226 and CD-ROM drive 230 may use, for example, an
integrated drive electronics (IDE) or serial advanced technology
attachment (SATA) interface. Super I/O (SIO) device 236 is
connected to SB/ICH 204.
[0096] An operating system runs on processing unit 206. The
operating system coordinates and provides control of various
components within the data processing system 200 in FIG. 2. As a
client, the operating system is a commercially available operating
system such as Microsoft.RTM. Windows 8.degree.. An object-oriented
programming system, such as the Java.TM. programming system, may
run in conjunction with the operating system and provides calls to
the operating system from Java.TM. programs or applications
executing on data processing system 200.
[0097] As a server, data processing system 200 may be, for example,
an IBM.RTM. eServer.TM. System p.RTM. computer system, running the
Advanced Interactive Executive) (AIX.RTM.) operating system or the
LINUX.RTM. operating system. Data processing system 200 may be a
symmetric multiprocessor (SMP) system including a plurality of
processors in processing unit 206. Alternatively, a single
processor system may be employed.
[0098] Instructions for the operating system, the object-oriented
programming system, and applications or programs are located on
storage devices, such as HDD 226, and are loaded into main memory
208 for execution by processing unit 206. The processes for
illustrative embodiments of the present invention are performed by
processing unit 206 using computer usable program code, which is
located in a memory such as, for example, main memory 208, ROM 224,
or in one or more peripheral devices 226 and 230, for example.
[0099] A bus system, such as bus 238 or bus 240 as shown in FIG. 2,
is comprised of one or more buses. Of course, the bus system may be
implemented using any type of communication fabric or architecture
that provides for a transfer of data between different components
or devices attached to the fabric or architecture. A communication
unit, such as modem 222 or network adapter 212 of FIG. 2, includes
one or more devices used to transmit and receive data. A memory may
be, for example, main memory 208, ROM 224, or a cache such as found
in NB/MCH 202 in FIG. 2.
[0100] Those of ordinary skill in the art will appreciate that the
hardware depicted in FIGS. 1 and 2 may vary depending on the
implementation. Other internal hardware or peripheral devices, such
as flash memory, equivalent non-volatile memory, or optical disk
drives and the like, may be used in addition to or in place of the
hardware depicted in FIGS. 1 and 2. Also, the processes of the
illustrative embodiments may be applied to a multiprocessor data
processing system, other than the SMP system mentioned previously,
without departing from the spirit and scope of the present
invention.
[0101] Moreover, the data processing system 200 may take the form
of any of a number of different data processing systems including
client computing devices, server computing devices, a tablet
computer, laptop computer, telephone or other communication device,
a personal digital assistant (PDA), or the like. In some
illustrative examples, data processing system 200 may be a portable
computing device that is configured with flash memory to provide
non-volatile memory for storing operating system files and/or
user-generated data, for example. Essentially, data processing
system 200 may be any known or later developed data processing
system without architectural limitation.
[0102] FIG. 3 illustrates a QA system pipeline, of a cognitive
system, for processing an input question in accordance with one
illustrative embodiment. The QA system pipeline of FIG. 3 may be
implemented, for example, as QA pipeline 108 of cognitive system
100 in FIG. 1. It should be appreciated that the stages of the QA
pipeline shown in FIG. 3 are implemented as one or more software
engines, components, or the like, which are configured with logic
for implementing the functionality attributed to the particular
stage. Each stage is implemented using one or more of such software
engines, components or the like. The software engines, components,
etc. are executed on one or more processors of one or more data
processing systems or devices and utilize or operate on data stored
in one or more data storage devices, memories, or the like, on one
or more of the data processing systems. The QA pipeline of FIG. 3
is augmented, for example, in one or more of the stages to
implement the improved mechanism of the illustrative embodiments
described hereafter, additional stages may be provided to implement
the improved mechanism, or separate logic from the pipeline 300 may
be provided for interfacing with the pipeline 300 and implementing
the improved functionality and operations of the illustrative
embodiments.
[0103] As shown in FIG. 3, the QA pipeline 300 comprises a
plurality of stages 310-380 through which the cognitive system
operates to analyze an input question and generate a final
response. In an initial question input stage 310, the QA pipeline
300 receives an input question that is presented in a natural
language format. That is, a user inputs, via a user interface, an
input question for which the user wishes to obtain an answer, e.g.,
"Who are Washington's closest advisors?" In response to receiving
the input question, the next stage of the QA pipeline 300, i.e. the
question and topic analysis stage 320, parses the input question
using natural language processing (NLP) techniques to extract major
features from the input question, and classify the major features
according to types, e.g., names, dates, or any of a plethora of
other defined topics. For example, in the example question above,
the term "who" may be associated with a topic for "persons"
indicating that the identity of a person is being sought,
"Washington" may be identified as a proper name of a person with
which the question is associated, "closest" may be identified as a
word indicative of proximity or relationship, and "advisors" may be
indicative of a noun or other language topic.
[0104] In addition, the extracted major features include key words
and phrases classified into question characteristics, such as the
focus of the question, the lexical answer type (LAT) of the
question, and the like. As referred to herein, a lexical answer
type (LAT) is a word in, or a word inferred from, the input
question that indicates the type of the answer, independent of
assigning semantics to that word. For example, in the question
"What maneuver was invented in the 1500s to speed up the game and
involves two pieces of the same color?," the LAT is the string
"maneuver." The focus of a question is the part of the question
that, if replaced by the answer, makes the question a standalone
statement. For example, in the question "What drug has been shown
to relieve the symptoms of ADD with relatively few side effects?,"
the focus is "drug" since if this word were replaced with the
answer, e.g., the answer "Adderall" can be used to replace the term
"drug" to generate the sentence "Adderall has been shown to relieve
the symptoms of ADD with relatively few side effects." The focus
often, but not always, contains the LAT. On the other hand, in many
cases it is not possible to infer a meaningful LAT from the
focus.
[0105] Referring again to FIG. 3, the identified major features are
then used during the question decomposition stage 330 to decompose
the question into one or more queries that are applied to the
corpora of data/information 345 in order to generate one or more
hypotheses. The queries are generated in any known or later
developed query language, such as the Structure Query Language
(SQL), or the like. The queries are applied to one or more
databases storing information about the electronic texts,
documents, articles, websites, and the like, that make up the
corpora of data/information 345. That is, these various sources
themselves, different collections of sources, and the like,
represent a different corpus 347 within the corpora 345. There may
be different corpora 347 defined for different collections of
documents based on various criteria depending upon the particular
implementation. For example, different corpora may be established
for different topics, subject matter categories, sources of
information, or the like. As one example, a first corpus may be
associated with healthcare documents while a second corpus may be
associated with financial documents. Alternatively, one corpus may
be documents published by the U.S. Department of Energy while
another corpus may be IBM Redbooks documents. Any collection of
content having some similar attribute may be considered to be a
corpus 347 within the corpora 345.
[0106] The queries are applied to one or more databases storing
information about the electronic texts, documents, articles,
websites, and the like, that make up the corpus of
data/information, e.g., the corpus of data 106 in FIG. 1. The
queries are applied to the corpus of data/information at the
hypothesis generation stage 340 to generate results identifying
potential hypotheses for answering the input question, which can
then be evaluated. That is, the application of the queries results
in the extraction of portions of the corpus of data/information
matching the criteria of the particular query. These portions of
the corpus are then analyzed and used, during the hypothesis
generation stage 340, to generate hypotheses for answering the
input question. These hypotheses are also referred to herein as
"candidate answers" for the input question. For any input question,
at this stage 340, there may be hundreds of hypotheses or candidate
answers generated that may need to be evaluated.
[0107] The QA pipeline 300, in stage 350, then performs a deep
analysis and comparison of the language of the input question and
the language of each hypothesis or "candidate answer," as well as
performs evidence scoring to evaluate the likelihood that the
particular hypothesis is a correct answer for the input question.
As mentioned above, this involves using a plurality of reasoning
algorithms, each performing a separate type of analysis of the
language of the input question and/or content of the corpus that
provides evidence in support of, or not in support of, the
hypothesis. Each reasoning algorithm generates a score based on the
analysis it performs which indicates a measure of relevance of the
individual portions of the corpus of data/information extracted by
application of the queries as well as a measure of the correctness
of the corresponding hypothesis, i.e. a measure of confidence in
the hypothesis. There are various ways of generating such scores
depending upon the particular analysis being performed. In
generally, however, these algorithms look for particular terms,
phrases, or patterns of text that are indicative of terms, phrases,
or patterns of interest and determine a degree of matching with
higher degrees of matching being given relatively higher scores
than lower degrees of matching.
[0108] Thus, for example, an algorithm may be configured to look
for the exact term from an input question or synonyms to that term
in the input question, e.g., the exact term or synonyms for the
term "movie," and generate a score based on a frequency of use of
these exact terms or synonyms. In such a case, exact matches will
be given the highest scores, while synonyms may be given lower
scores based on a relative ranking of the synonyms as may be
specified by a subject matter expert (person with knowledge of the
particular domain and terminology used) or automatically determined
from frequency of use of the synonym in the corpus corresponding to
the domain. Thus, for example, an exact match of the term "movie"
in content of the corpus (also referred to as evidence, or evidence
passages) is given a highest score. A synonym of movie, such as
"motion picture" may be given a lower score but still higher than a
synonym of the type "film" or "moving picture show." Instances of
the exact matches and synonyms for each evidence passage may be
compiled and used in a quantitative function to generate a score
for the degree of matching of the evidence passage to the input
question.
[0109] Thus, for example, a hypothesis or candidate answer to the
input question of "What was the first movie?" is "The Horse in
Motion." If the evidence passage contains the statements "The first
motion picture ever made was `The Horse in Motion` in 1878 by
Eadweard Muybridge. It was a movie of a horse running," and the
algorithm is looking for exact matches or synonyms to the focus of
the input question, i.e. "movie," then an exact match of "movie" is
found in the second sentence of the evidence passage and a highly
scored synonym to "movie," i.e. "motion picture," is found in the
first sentence of the evidence passage. This may be combined with
further analysis of the evidence passage to identify that the text
of the candidate answer is present in the evidence passage as well,
i.e. "The Horse in Motion." These factors may be combined to give
this evidence passage a relatively high score as supporting
evidence for the candidate answer "The Horse in Motion" being a
correct answer.
[0110] It should be appreciated that this is just one simple
example of how scoring can be performed. Many other algorithms of
various complexity may be used to generate scores for candidate
answers and evidence without departing from the spirit and scope of
the present invention.
[0111] In the synthesis stage 360, the large number of scores
generated by the various reasoning algorithms are synthesized into
confidence scores or confidence measures for the various
hypotheses. This process involves applying weights to the various
scores, where the weights have been determined through training of
the statistical model employed by the QA pipeline 300 and/or
dynamically updated. For example, the weights for scores generated
by algorithms that identify exactly matching terms and synonym may
be set relatively higher than other algorithms that are evaluating
publication dates for evidence passages. The weights themselves may
be specified by subject matter experts or learned through machine
learning processes that evaluate the significance of
characteristics evidence passages and their relative importance to
overall candidate answer generation.
[0112] The weighted scores are processed in accordance with a
statistical model generated through training of the QA pipeline 300
that identifies a manner by which these scores may be combined to
generate a confidence score or measure for the individual
hypotheses or candidate answers. This confidence score or measure
summarizes the level of confidence that the QA pipeline 300 has
about the evidence that the candidate answer is inferred by the
input question, i.e. that the candidate answer is the correct
answer for the input question.
[0113] The resulting confidence scores or measures are processed by
a final confidence merging and ranking stage 370 which compares the
confidence scores and measures to each other, compares them against
predetermined thresholds, or performs any other analysis on the
confidence scores to determine which hypotheses/candidate answers
are the most likely to be the correct answer to the input question.
The hypotheses/candidate answers are ranked according to these
comparisons to generate a ranked listing of hypotheses/candidate
answers (hereafter simply referred to as "candidate answers"). From
the ranked listing of candidate answers, at stage 380, a final
answer and confidence score, or final set of candidate answers and
confidence scores, are generated and output to the submitter of the
original input question via a graphical user interface or other
mechanism for outputting information.
[0114] As shown in FIG. 3, in accordance with one illustrative
embodiment, the QA system pipeline 300 operates in conjunction with
a temporal relevance evaluation engine 390 which provides logic and
functionality for evaluating the temporal characteristics of
documents, portions of content within documents, input questions,
tokens of candidate answers, and scoring candidate answers based on
the correspondence of these temporal characteristics of the input
question and tokens associated with the candidate answers. It
should be appreciated that while FIG. 3 shows the temporal
relevance evaluation engine 390 as being a separate element from
the QA system pipeline 300, in some illustrative embodiments, one
or more of the elements of the temporal relevance evaluation engine
390, or the engine 390 as a whole, may be integrated into the QA
system pipeline 300 as an additional stage or as additional logic
added to one or more of the individual stages 310-380 of the QA
system pipeline 300, without departing from the spirit and scope of
the present invention. For example, in one illustrative embodiment,
the input question temporal foci evaluation logic 396 may be
integrated into the question and topic analysis stage logic 320,
the candidate answer temporal foci logic 398 may be integrated into
the hypothesis generation stage logic 340, and the candidate answer
temporal relevance scoring logic 399 may be integrated into the
hypothesis and evidence scoring stage logic 350.
[0115] The document temporal expression datetime normalization
logic 392 and document token temporal foci logic 394, which perform
the operations and provide the logic as discussed above with regard
to elements 122 and 124 in FIG. 1, may operate as part of a
pre-processing operation that operates on the contents of documents
of the corpus or corpora 345, 347 prior to the QA system pipeline
300 utilizing the content of these documents to answer questions
submitted to the QA system pipeline 300. Thus, as part of an
ingestion operation, in addition to the various analysis and
annotation performed by the mechanisms of the cognitive system and
QA system discussed above to generate an in-memory representation
of the documents of the corpus or corpora 345, 347, the additional
processing discussed previously with regard to elements 122 and 124
may be performed by the document temporal expression datetime
normalization logic 392 and document token temporal foci logic 394
to associate with tokens in the content datetimes based on
evaluations of document relevance datetimes, definite temporal
expressions, and identifying a closest definite temporal expression
to the token that is used to associate its temporal focus or foci
with the token. Thus, each token in the content of a document will
have its own associated temporal focus or foci. Alternatively, as
mentioned above, these processes may be performed during realtime
in response to an input question being received by the QA system
pipeline 300 and may be done with regard to a specific document
associated with a candidate answer identified through the
processing of the QA system pipeline 300.
[0116] In addition, in response to the QA system pipeline 300
receiving an input question and performing its initial processing
of the input question via stages 310 and 320, the input question,
or the results of the processing of the input question via stage
320, is provided to the input question temporal foci evaluation
logic 396 which performs operations similar to that described above
with regard to element 126 of FIG. 1. That is, the input question
temporal foci evaluation logic 396 operates to determine one or
more temporal foci of the input natural language question. As
discussed previously, this may involve identifying one or more
temporal foci using the current datetime as the relevant contextual
datetime for the input question and identifying and normalizing all
definite temporal expressions in the question with respect to this
relevant contextual datetime. This identification and normalization
may comprise determining if the question itself specifies a
particular datetime and if not, whether the input question
comprises a definite temporal expression that is a relative
definite temporal expression that can be evaluated relative to the
relevant contextual datetime of the input question. If there are no
definite temporal expressions in the input question, the verb tense
of the input question may be evaluated to determine if the input
question is presented in present tense in which case the current
datetime may be selected as the temporal focus of the input
question; otherwise no datetime is selected.
[0117] These operations of input question temporal foci evaluation
logic 396 may be performed in parallel with the operations of
stages 320-340 which perform their operations for generating a set
of candidate answers to the input question. The candidate answers
generated, for example by hypothesis generation stage logic 340 of
the QA system pipeline 300, may be further evaluated by the
candidate answer temporal foci logic 398 and candidate answer
temporal relevance scoring logic 399 working in conjunction with
the QA system pipeline 300.
[0118] As noted above, either previously, through operation of an
ingestion process in which the above operations are performed to
associate datetimes with tokens in the documents, or as part of the
processing of the input question, the tokens that make up the
candidate answers generated by the hypothesis generation stage
logic 340 are used by the candidate answer temporal foci logic 398
to generate one or more temporal foci of the corresponding
candidate answer. That is, the candidate answer temporal foci logic
398 takes the candidate answers generated by the hypothesis
generation stage logic 340 and identifies the datetimes associated
with the tokens that make up the candidate answer. The temporal
foci of the tokens of the candidate answer are then evaluated to
generate a single contextual temporal focus of the candidate
answer, such as by way of performing a union of the temporal foci
of the various tokens of the candidate answer or performing a more
complex relationship evaluation of the foci of the tokens, as
discussed previously. The result is that each candidate answer
identified by the hypothesis generation stage logic 340 is
associated with a corresponding contextual temporal focus by the
candidate answer temporal foci logic 398. Moreover, the input
question has its own set of one or more temporal foci as determined
by the input question temporal foci evaluation logic 396.
[0119] Having determined a temporal focus or foci for the input
question and a contextual temporal focus for each of the candidate
answers, the candidate answers are then scored by the candidate
temporal relevance scoring logic 399 according to the temporal
relevance of the candidate answer with respect to the input
question. That is, a score is attributed to each candidate answer
based on the relationship of its corresponding contextual temporal
focus with the one or more temporal foci of the input question. A
range of scores may be established such that a highest score
indicates an exact match between the contextual temporal focus of
the candidate answer and the temporal focus or foci of the input
question, and a lowest score indicates a complete mis-match between
the contextual temporal focus of the candidate answer and the
temporal focus or foci of the input question. A complete mis-match
may be measured in many different ways, e.g., being outside a
datetime range of the temporal focus or foci of the input
question.
[0120] As noted above, in one illustrative embodiment, the
candidate answer may be given a first score, e.g., a "1", if there
is a datetime in the temporal focus or foci of the input question
which overlaps the datetime in the temporal focus of the candidate
answer (contextual temporal focus). Otherwise, if there is no
overlap of this nature, then the candidate answer may be given a
second score, e.g., "0". In other illustrative embodiments, a more
complex scoring may be used by the candidate answer temporal
relevance scoring logic 399 which is based on how close the
temporal foci in the input question are to the contextual temporal
focus of the candidate answer, such that a range of scores between
the first and second scores may be assigned to a candidate answer.
Thus, for example, if the contextual temporal focus is only a month
or two before the timeframe represented by the temporal focus of
the input question, then the score for the candidate answer may be
relatively higher than a candidate answer whose contextual temporal
focus is many years before or after the temporal focus of the input
question.
[0121] Thus, the candidate answer temporal relevance scoring logic
399 generates a temporal relevance scoring of the candidate answers
based on each candidate answers's temporal relevance to the input
question. These temporal relevance scores may be provided to the
hypothesis and evidence scoring stage logic 350 for use in a more
complex scoring of candidate answers that performs scoring based on
evidence passages in the corpus or corpora 345, 347 as well as the
temporal relevance scoring. In some illustrative embodiments
candidate answers that are determined by the candidate answer
temporal relevance scoring logic 399 to have no temporal relevance
may in fact be eliminated from further evaluation by the logic of
the QA system pipeline 300. Thus, the temporal scores associated
with candidate answers generated by the hypothesis generation stage
logic 340 may be used to prune the set of candidate answers prior
to further evidential evaluation by the hypothesis and evidence
scoring stage logic 350. In other illustrative embodiments, the
temporal scores associated with candidate answers may be used as a
weight to be applied to the evidence scores generated by the
hypothesis and evidence scoring stage logic 350 such that candidate
answers that are determined to be more temporally relevant to the
input question are weighted more heavily than candidate answers
that are not as temporally relevant. Alternatively, in other
illustrative embodiments, the temporal scores of candidate answers
may be simply another scoring factor that is weighted according to
its determined relative influence on the correctness of candidate
answers, which is then combined with the weighted scores of other
evidence from the corpus or corpora 345, 347, to generate an
overall confidence score for the candidate answers. Any mechanism
for integrating the temporal relevance scoring into an overall
scoring of candidate answers may be used without departing from the
spirit and scope of the present invention.
[0122] Thereafter, the operation of the QA system pipeline 300 is
essentially the same as already discussed above. That is, the
synthesis stage logic 360, final confidence merging and ranking
stage logic 370, and final answer and confidence stage logic 380
operate to rank the candidate answers according to their confidence
scores, which includes the temporal scoring discussed above, and
select one or more final answers to be returned as answers to the
input question. Thus, temporal relevance is evaluated using the
mechanisms of the illustrative embodiments to provide more accurate
evaluations of candidate answers.
[0123] FIG. 4 is a flowchart outlining an example operation for
ingesting documents of a corpus and associating datetimes with
tokens in the documents in accordance with one illustrative
embodiment. The operation outlined in FIG. 4 is shown for an
embodiment in which the operations are part of a pre-processing or
ingestion of documents of a corpus for use with a cognitive system.
That is, the operations in FIG. 4 may be performed prior to
handling a cognitive operation request, such as the input of a
natural language question or search query, for example. It should
be appreciated that minor adjustments to the operation shown in
FIG. 4 may be made to perform similar operations during runtime
processing of a cognitive operation request, such as with regard to
a document or portion of content associated with a potential result
of the cognitive operation, e.g., a candidate answer to an input
question or a candidate search result for a search query.
[0124] The operation outlined in FIG. 4 may be performed, for
example, by the document temporal expression datetime normalization
logic 122, 392 and document token temporal foci logic 124, 394, for
example. As shown in FIG. 4, the operation starts by receiving a
corpus of documents for ingestion (step 410). A document relevance
datetime for each of the documents is identified (step 420), such
as by processing metadata associated with each of the documents. As
noted above, this document relevance datetime may be a publication
date/time, creation date/time, ingestion date/time, or the like.
Definite temporal expressions in content of the documents are
identified (step 430) and a temporal focus of foci is associated
with each of the identified definite temporal expressions based on
analysis of expressions and corresponding document relevance
datetimes (step 440). Thereafter, each of the tokens in the
documents are identified and associated with a closest definite
temporal expression, if any, such that the corresponding temporal
focus or foci may be associated with the token (step 450).
Alternatively, the document relevance datetime may be used to
associate a temporal focus or foci with the tokens if there is no
closest definite temporal expression. The temporal foci associated
with the tokens is stored in association with the tokens in the
in-memory representation of the documents of the corpus for further
use by the cognitive system when performing cognitive operations,
such as question answering, cognitive searching of the corpus, or
the like.
[0125] FIG. 5 is a flowchart outlining an example operation for
evaluating candidate answers to an input question based on a
temporal focus of the input question and contextual temporal foci
of candidate answers in accordance with one illustrative
embodiment. While the flowchart shown in FIG. 5 is for processing
an input natural language question, it should be appreciated that
modifications to the operation of FIG. 5 may be made to apply
similar operations to other types of cognitive system inputs to
perform other types of cognitive operations, such as cognitive
searches of natural language content provided in documents of a
corpus based on a search query, cognitive evaluations of patient
medical records by a cognitive system employing a patient registry
which is searched in accordance with a cognitive request, e.g.,
diagnosis request, patient evaluation with regard to specific
criteria, or the like.
[0126] As shown in FIG. 5, the operation starts by receiving an
input question (step 510) and one or more temporal foci of the
input question are identified (step 520). This may be performed,
for example, by the input question temporal foci evaluation logic
126, 396 in FIGS. 1 and 3. The input question is processed by the
QA system to generate one or more candidate answers (step 530). A
contextual temporal focus is determined for each candidate answer
based on the temporal foci of the tokens associated with the
candidate answer (step 540). This may be performed, for example, by
the candidate answer temporal foci logic 128, 398.
[0127] The candidate answers are then each scored according to
results of a comparison of the contextual temporal focus of the
candidate answer and the temporal focus or foci of the input
question (step 550). This temporal scoring may be performed, for
example, by the candidate answer temporal relevance scoring logic
129, 399. The temporal scoring is then combined with the scoring of
candidate answers based on evidential basis to generate a
confidence score for each candidate answer (step 560). The
candidate answers are then ranked according to the confidence
scores of the candidate answers, including the temporal scoring,
and one or more final answers to the input question are selected
(step 570). The selected final answer(s) are then returned as
answers to the input question (step 580). The operation then
terminates.
[0128] It should be appreciated that while the above illustrative
embodiments have been described in the context of a QA system
answering an input question, the illustrative embodiments are not
limited to such. Rather the illustrative embodiments may be
implemented in any cognitive system that processes requests based
on documents in a corpus of documents using cognitive logic
processes. For example, the illustrative embodiments may be
utilized in a cognitive search engine where, rather than an input
question, a search query may be input and the search query may be
processed to identify temporal foci with search results being
returned and evaluated by the mechanisms of the illustrative
embodiments to identify a contextual temporal focus of the search
results, which are then scored and ranked in accordance with the
mechanisms of the illustrative embodiments as described above.
Other cognitive systems based on natural language processing of
documents or other content may also be augmented with the
mechanisms of the illustrative embodiments to evaluate portions of
text with regard to a temporal relevance to a particular input,
e.g., patient electronic record evaluation systems, cognitive law
enforcement systems, or the like. Any cognitive system that
analyzes textual content may be augmented to include the temporal
relevance evaluation logic of the illustrative embodiments to
evaluate the relevance of one portion of text to another portion of
text, without departing from the spirit and scope of the present
invention.
[0129] As discussed above, while the example embodiments set forth
in the Figures and described herein a primarily directed to the
answering of natural language input questions using a corpus or
corpora of natural language documents, the illustrative embodiments
are not limited to such and any cognitive system performing
cognitive operations may make use of the mechanisms of the
illustrative embodiments to determine a temporal relevance of
results to an initial request. For example, with regard to a
cognitive system that operates as a cognitive search engine, rather
than an input natural language question being processed, an input
of a natural language search query may be received and processed by
the cognitive search engine so as to return search results that are
most relevant to the search query. The operations of the
illustrative embodiments may be used to process the input search
query to identify a temporal focus or foci of the search query and
then compare contextual temporal foci of potential search results,
found by performing keyword searching and the like, with the
temporal foci of the input search query. The ranking of potential
search results may then be made based on a scoring of the potential
search results both with regard to a degree of matching of the
keywords of the search query as well as the temporal relevance of
the search results to the search query as determined from the
comparison of temporal foci and contextual temporal foci.
[0130] For example, the search query may be of the type "news
stories about bankruptcy filings in the last month." The search
query may be analyzed using the mechanisms of the illustrative
embodiments to determine that the focus or foci of the input search
query is a range of datetimes from the current datetime back one
month. The search query may be evaluated to identify search results
matching the keywords or criteria of the search query, e.g.,
portions of content that are new stories that discuss bankruptcy
filings. The temporal focus or foci of the input search query may
be evaluated against the contextual temporal focus of potential
search results to score them according to their temporal relevance
to the input search query, rank them, and return a ranked set of
search results.
[0131] In other cognitive systems, such as a patient medical
records evaluation system, the input query may represent a
particular request to diagnose the patient, find patients with
certain types of diagnoses, find patients with certain
characteristics, and the like, that are tied to a temporal focus or
foci. For example, the initial request may be posed as a search
request or as an input question and may specify the criteria for
finding results and the temporal focus or foci of the request. An
example may be of the type, "what patients had diagnosis in the
last year that indicated diabetes?" Such a request may be processed
using the mechanisms of the illustrative embodiments in the manner
previously described above.
[0132] As noted above, it should be appreciated that the
illustrative embodiments may take the form of an entirely hardware
embodiment, an entirely software embodiment or an embodiment
containing both hardware and software elements. In one example
embodiment, the mechanisms of the illustrative embodiments are
implemented in software or program code, which includes but is not
limited to firmware, resident software, microcode, etc.
[0133] A data processing system suitable for storing and/or
executing program code will include at least one processor coupled
directly or indirectly to memory elements through a system bus. The
memory elements can include local memory employed during actual
execution of the program code, bulk storage, and cache memories
which provide temporary storage of at least some program code in
order to reduce the number of times code must be retrieved from
bulk storage during execution.
[0134] Input/output or I/O devices (including but not limited to
keyboards, displays, pointing devices, etc.) can be coupled to the
system either directly or through intervening I/O controllers.
Network adapters may also be coupled to the system to enable the
data processing system to become coupled to other data processing
systems or remote printers or storage devices through intervening
private or public networks. Modems, cable modems and Ethernet cards
are just a few of the currently available types of network
adapters.
[0135] The description of the present invention has been presented
for purposes of illustration and description, and is not intended
to be exhaustive or limited to the invention in the form disclosed.
Many modifications and variations will be apparent to those of
ordinary skill in the art without departing from the scope and
spirit of the described embodiments. The embodiment was chosen and
described in order to best explain the principles of the invention,
the practical application, and to enable others of ordinary skill
in the art to understand the invention for various embodiments with
various modifications as are suited to the particular use
contemplated. The terminology used herein was chosen to best
explain the principles of the embodiments, the practical
application or technical improvement over technologies found in the
marketplace, or to enable others of ordinary skill in the art to
understand the embodiments disclosed herein.
* * * * *